Shulou(Shulou.com)06/01 Report--

As an application developer, database applications have been very widespread. You may have used relational data, such as MySQL, PostgreSQL, or document storage, such as MongoDB, or key-value databases, such as Redis. Each database has its strengths, and you may also be considering using distributed databases, such as Cassandra, to solve the task at hand.

The use of these data products is not to replace the original data products, but to provide more choices for different application scenarios. NoSQL stands for choosing the right solution to deal with the right business scenario.

In the course "introduction to Cassandra," we will discuss the main reasons for the transition from a relational database to Cassandra, as well as the basic features of Cassandra. At the end of this chapter, you should learn something:

Characteristics of RDBMS

Is RDBMS suitable for big data?

The third paradigm is not scalable.

Sharding is a nightmare.

Highly available.. It's not real.

Summary of shortcomings

Course summary

Let's start with a look at relational databases:

Characteristics of RDBMS

RDBMS is suitable for medium-sized data and works well on a single machine, such as MySQL and PostgreSQL.

It is well supported for hundreds of concurrent users.

ACID supports well

Is RDBMS suitable for big data?

For big data, it is necessary to scale horizontally. MySQL's master/slave mode will cause ACID (A: atomicity, C: consistency, I: isolation, D: persistence) to cease to exist.

The third paradigm is not scalable (no redundancy)

Due to the complexity of the query, and the user needs to respond quickly at the same time, because the user is impatient, the data must be de-styled.

Sharding is a nightmare.

The data is located in every shard

Join and aggregation difficulties

Need to be anti-stereotyped

The query needs to use shard rules or routes to hit shard

Adding shard requires manual migration of data

Highly available.. It's not real.

Master is a single point of failure

Multiple data centers are not supported

Summary of shortcomings

Horizontal expansion is a headache.

ACID is best locally, and there is a consistency problem with multiple computers.

Re-sharding requires manual data migration

Anti-normalization is often needed for performance.

High availability is complex and requires additional operation

Course summary

Since RDBMS has the above shortcomings, we need to address them:

Strong consistency is unrealistic: So, give him up

Re-sharding is difficult: So, we need to do it automatically

Master failover:So, we should not use master/slave mode

Data distribution and aggregation no good:So, for real-time query performance, need to be de-normalized in order to make the query always hit on one machine

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.